Vacuum State and Zero-Point Energy

Field quantization hands quantum optics a startling inheritance: every mode of the electromagnetic field, labelled by wavevector $\mathbf{k}$ and polarization $\lambda$ , is a harmonic oscillator with Hamiltonian $\hat{H}=\hbar\omega(\hat{a}^\dagger\hat{a}+\tfrac12)$ . The two preceding articles built the ladder operators and the Fock basis; this one asks what sits at the bottom of the ladder. The answer is the vacuum state $|0\rangle$ — the unique state annihilated by every lowering operator, $\hat{a}_{\mathbf{k},\lambda}|0\rangle=0$ , the true ground state of the field with not one photon in any mode. And yet it is not the nothing classical physics would call empty space. The tell is the stubborn $\tfrac12\hbar\omega$ the oscillator drags into its ground state. It survives because $[\hat{a},\hat{a}^\dagger]=1$ forbids the two quadratures from being sharp at once, just as $[\hat{x},\hat{p}]=i\hbar$ forbids a particle from lying still at a potential minimum. So the mean field vanishes, $\langle 0|\hat{\mathbf{E}}|0\rangle=0$ , but its variance does not: $\langle 0|\hat{\mathbf{E}}^2|0\rangle\neq0$ . Those vacuum fluctuations are the physical body of the half-quantum, no mere bookkeeping — they knock excited atoms into emitting, shift atomic levels, and pull uncharged mirrors together. Summed over the field's infinitely many modes the zero-point energy diverges, $E_0=\sum_{\mathbf{k},\lambda}\tfrac12\hbar\omega_{\mathbf{k}}\to\infty$ — not a catastrophe but a signpost: only changes in vacuum energy, when mirrors, atoms, or dielectrics reshape which modes are allowed, are observable. Coherent states are displaced vacua, squeezed states have reshaped fluctuations, Fock states are $\hat{a}^\dagger$ on $|0\rangle$ : the vacuum is the reference state on which all of quantum optics is drawn.

Key takeaways

The vacuum $|0\rangle$ is the ground state of the quantized field: zero photons in every mode, yet an energy $\tfrac12\hbar\omega$ per mode that cannot be removed.
Because $[\hat{a},\hat{a}^\dagger]=1$ , the two quadratures cannot both vanish; the vacuum carries irreducible fluctuations $\langle\Delta\hat{X}_1^2\rangle=\langle\Delta\hat{X}_2^2\rangle=\tfrac14$ that saturate the uncertainty relation.
The mean field is zero, $\langle 0|\hat{\mathbf{E}}|0\rangle=0$ , but the variance is not — the second moment drives spontaneous emission, the Lamb shift, and optical shot noise.
Summed over all modes the zero-point energy diverges; only differences (the Casimir force, level shifts) are physical, so the divergence never reaches an observable.
The vacuum is the origin of phase space against which coherent, squeezed, and Fock states are defined, and it sets the standard quantum limit for optical measurement.
Key idea: a quantized field mode is a harmonic oscillator whose ground state has energy $\tfrac12\hbar\omega$ and irreducible fluctuations, because $[\hat{a},\hat{a}^\dagger]=1$ forbids both quadratures from vanishing together.

The core insight

A quantized field mode is a harmonic oscillator; its ground state has energy $\tfrac12\hbar\omega$ and irreducible fluctuations because $[\hat{a},\hat{a}^\dagger]=1$ forbids both quadratures from vanishing at once.

Where this lives today

The vacuum has moved from a conceptual worry to an engineering resource. In gravitational-wave detection, vacuum fluctuations entering the interferometer's dark port set the shot-noise floor that limits how faint a strain LIGO and Virgo can hear; since 2019 both inject squeezed vacuum through that port, buying real high-frequency sensitivity without adding a watt of laser power. At the opposite scale, the Casimir force is now a design constraint: in nanoelectromechanical systems the same $1/d^4$ attraction that is imperceptible at microns causes stiction that welds moving parts together at nanometre gaps, and metrology groups measure it to hunt for short-range forces beyond the standard model. Cavity and circuit QED engineer the vacuum directly — enhancing or suppressing spontaneous emission by tailoring an emitter's mode density (the Purcell effect) — and the dynamical Casimir effect, photons conjured from the vacuum by relativistically moved mirrors, has been seen in superconducting circuits. The zero-point field is now routinely measured, reshaped, and put to work.

The physical picture

Picture a single field mode as a quantum pendulum. Even in its lowest state it refuses to hang still, because holding it exactly at rest would demand a simultaneously sharp position and momentum — forbidden. The field's "position" and "momentum" are its two quadratures $\hat{X}_1$ and $\hat{X}_2$ , and the vacuum is the state of least combined jitter: in phase space it sits at the origin, both quadrature means zero so the field is dark on average, yet the disk has a finite radius set by the variance $\tfrac14$ in each direction — no photodetector clicks from empty space, but the disk is not a point. That finite radius makes the "empty" vacuum dynamically alive. The same fluctuations that stop the pendulum from resting kick an excited atom into radiating with no photons present (spontaneous emission), smear the electron's Coulomb interaction to split hydrogen's levels (the Lamb shift), and stamp a noise floor onto every optical measurement (shot noise). Because $|0\rangle$ is an eigenstate of $\hat{n}$ with eigenvalue zero, the photon number is perfectly defined while the phase is completely undefined — a number state maximising quadrature uncertainty at the origin.

Figure

The vacuum as a phase-space fuzzball at the origin: both quadrature means vanish, ⟨X₁⟩ = ⟨X₂⟩ = 0, yet each has irreducible variance ⟨ΔX₁²⟩ = ⟨ΔX₂²⟩ = 1/4, giving the disk radius 1/2 — a coherent state is the same disk pushed off the origin by α

The divergence in $\sum\tfrac12\hbar\omega$ is a symptom of counting infinitely many modes, not a paradox: physics only ever measures how this energy changes when you insert mirrors, atoms, or dielectrics. Hendrik Casimir's insight was that two conducting plates forbid the long-wavelength modes that would otherwise fit between them, lowering the vacuum energy in the gap below that outside, so the plates are pushed together by a pressure built from nothing but $\hbar$ , $c$ , and $d$ — the cleanest proof that what is real is not the vacuum energy's infinite absolute value but its response to boundaries.

Figure

Two conducting plates at separation d admit only standing-wave modes that fit between them, excluding the long-wavelength vacuum modes outside; the leftover pressure P = π²ħc / 240 d⁴ ≈ 1.3 mPa at d = 1 μm climbs as 1/d⁴ to about one atmosphere near d = 10 nm

The core equations

The vacuum is defined not by what it contains but by what destroys it: it is the one state every annihilation operator sends to zero. This single line fixes the whole state, since something emptied by lowering in every mode can lose nothing further.

\hat{a}_{\mathbf{k},\lambda}|0\rangle=0\quad\text{for all }\mathbf{k},\lambda \qquad (1)

where: $\hat{a}_{\mathbf{k},\lambda}$ — annihilation operator for the mode of wavevector $\mathbf{k}$ and polarization $\lambda$ ; $|0\rangle$ — the multimode vacuum, joint ground state of every mode. Because $\hat{n}|0\rangle=0$ , the photon number is exactly zero everywhere. Each mode contributes an oscillator Hamiltonian, and the field Hamiltonian is their sum. The $+\tfrac12$ inside every bracket is the term that refuses to go away.

\hat{H}=\sum_{\mathbf{k},\lambda}\hbar\omega_{\mathbf{k}}\left(\hat{a}^\dagger_{\mathbf{k},\lambda}\hat{a}_{\mathbf{k},\lambda}+\tfrac12\right) \qquad (2)

where: $\hbar\omega_{\mathbf{k}}$ — energy quantum of the mode; $\hat{a}^\dagger\hat{a}=\hat{n}$ — photon-number operator; $\tfrac12$ — zero-point contribution per mode, independent of occupation. Taking the vacuum expectation of $\hat{H}$ kills every $\hat{n}$ term and leaves only the half-quanta, summed over all modes. The result is the total zero-point energy, and it is formally infinite.

E_0=\langle 0|\hat{H}|0\rangle=\sum_{\mathbf{k},\lambda}\tfrac12\hbar\omega_{\mathbf{k}} \qquad (3)

where: $E_0$ — total vacuum energy. The sum runs over unbounded $\omega_{\mathbf{k}}$ , so $E_0$ diverges; only differences of $E_0$ under changed boundaries are observable. The reason the half-quantum cannot be dropped is a single commutator, the one underlying the whole ladder. Its non-vanishing forbids a zero-energy ground state.

[\hat{a},\hat{a}^\dagger]=1 \qquad (4)

where: $\hat{a},\hat{a}^\dagger$ — single-mode ladder operators. Because $\hat{a}\hat{a}^\dagger=\hat{a}^\dagger\hat{a}+1$ , symmetric ordering leaves an irreducible $\tfrac12\hbar\omega$ no state can remove. The field's two measurable amplitudes are the quadratures — the cosine and sine components of the oscillation. They are the field analogues of a particle's position and momentum, the coordinates of phase space.

\hat{X}_1=\tfrac12(\hat{a}+\hat{a}^\dagger),\qquad \hat{X}_2=\tfrac{1}{2i}(\hat{a}-\hat{a}^\dagger) \qquad (5)

where: $\hat{X}_1$ — amplitude (cosine) quadrature; $\hat{X}_2$ — phase (sine) quadrature. They obey $[\hat{X}_1,\hat{X}_2]=\tfrac{i}{2}$ , forcing $\Delta\hat{X}_1\,\Delta\hat{X}_2\ge\tfrac14$ . Evaluated in the vacuum, both quadratures have zero mean but equal, non-zero variance. Their product sits exactly on the uncertainty floor — a minimum-uncertainty state shared symmetrically between them.

\langle 0|\Delta\hat{X}_1^2|0\rangle=\langle 0|\Delta\hat{X}_2^2|0\rangle=\tfrac14 \qquad (6)

where: $\Delta\hat{X}_i^2$ — variance of quadrature $i$ about its zero mean. Each contributes $\tfrac14\hbar\omega$ , reproducing $\tfrac12\hbar\omega$ ; squeezing trades variance between them without changing the product. Turning the quadratures back into the real electric field, the vacuum mean vanishes while the mean square does not. That finite second moment is the mathematical home of the fluctuations that do all the physics.

\langle 0|\hat{\mathbf{E}}|0\rangle=0,\qquad \langle 0|\hat{\mathbf{E}}^2|0\rangle=\sum_{\mathbf{k},\lambda}\frac{\hbar\omega_{\mathbf{k}}}{2\varepsilon_0 V} \qquad (7)

where: $\hat{\mathbf{E}}$ — electric-field operator; $V$ — quantization volume; $\varepsilon_0$ — vacuum permittivity. The per-mode term is the square of the single-photon field amplitude $\mathcal{E}_{\mathbf{k}}=\sqrt{\hbar\omega_{\mathbf{k}}/2\varepsilon_0 V}$ ; when two mirrors reshape the modes, the finite change in this energy is the measurable Casimir force derived below.

Derivation AWhy the ground state keeps half a quantum

The goal is to show, from nothing but $[\hat{a},\hat{a}^\dagger]=1$ and $\hat{a}|0\rangle=0$ , that a single field mode cannot have zero ground-state energy but must carry exactly $\tfrac12\hbar\omega$ — and that this half-quantum is literally the sum of the two quadrature fluctuations, split evenly like a field-theory equipartition. We rewrite the Hamiltonian in quadratures, evaluate each variance in the vacuum, and add them, turning "energy of empty space" into "energy of the irreducible jitter".

Step 1. Express the single-mode Hamiltonian in quadratures: substituting $\hat{X}_1,\hat{X}_2$ and using $[\hat{a},\hat{a}^\dagger]=1$ gives $\hat{X}_1^2+\hat{X}_2^2=\hat{a}^\dagger\hat{a}+\tfrac12$ , so $\hat{H}=\hbar\omega(\hat{X}_1^2+\hat{X}_2^2)$ . Why: this mirrors $H=\tfrac{1}{2m}p^2+\tfrac12 m\omega^2 x^2$ — energy quadratic in the two quadratures, so the ground-state energy is set by how small their fluctuations can jointly be.

Step 2. Evaluate the variances: since $\hat{a}|0\rangle=0$ the means vanish, so each variance is the raw second moment; expanding $\hat{X}_1^2=\tfrac14(\hat{a}^2+\hat{a}^{\dagger2}+\hat{a}\hat{a}^\dagger+\hat{a}^\dagger\hat{a})$ in $|0\rangle$ , only $\hat{a}\hat{a}^\dagger=\hat{a}^\dagger\hat{a}+1$ survives, giving $\langle 0|\hat{X}_1^2|0\rangle=\langle 0|\hat{X}_2^2|0\rangle=\tfrac14$ . Why: the lone surviving $+1$ is the commutator itself — the fluctuation is the fingerprint of $[\hat{a},\hat{a}^\dagger]=1$ , and with zero mean the whole ground-state energy is fluctuation energy.

Step 3. Add them: $\langle 0|\hat{H}|0\rangle=\hbar\omega(\tfrac14+\tfrac14)=\tfrac12\hbar\omega$ . Why: were the ground-state energy zero, both variances would vanish, violating $\Delta\hat{X}_1\,\Delta\hat{X}_2\ge\tfrac14$ ; the half-quantum is the minimum energy the uncertainty relation allows. *Verified: reproduces the harmonic-oscillator ground-state energy $\tfrac12\hbar\omega$ and $\langle\hat{X}_i^2\rangle=\tfrac14$ (Loudon ch. 4).*

Go deeper · Derivation B · 1 minThe vacuum field variance and its ultraviolet divergence

Derivation BThe vacuum field variance and its ultraviolet divergence

The goal is to turn "the vacuum has fluctuations" into the concrete spectral statement $\rho_0(\omega)=\hbar\omega^3/2\pi^2c^3$ — the zero-point energy per unit volume per unit frequency — and to see why the integrated energy diverges. We expand the field into modes, take the vacuum expectation of its square, convert the sum to a frequency integral, and read off both the spectral density and the ultraviolet catastrophe inside it.

Step 1. Write the field as a mode sum, $\hat{\mathbf{E}}(\mathbf{r})=\sum_{\mathbf{k},\lambda}\mathcal{E}_{\mathbf{k}}\,\boldsymbol{\epsilon}_{\mathbf{k},\lambda}(\hat{a}_{\mathbf{k},\lambda}e^{i\mathbf{k}\cdot\mathbf{r}}+\text{h.c.})$ with $\mathcal{E}_{\mathbf{k}}=\sqrt{\hbar\omega_{\mathbf{k}}/2\varepsilon_0 V}$ . Why: $\mathcal{E}_{\mathbf{k}}$ is fixed by demanding one photon carry energy $\hbar\omega$ in volume $V$ , tying field strength to energy.

Step 2. Take the vacuum expectation of $\hat{\mathbf{E}}^2$ : cross terms and the $\hat{a}^2$ , $\hat{a}^{\dagger2}$ pieces vanish, only $\hat{a}\hat{a}^\dagger$ per mode survives, giving $\langle 0|\hat{\mathbf{E}}^2|0\rangle=\sum_{\mathbf{k},\lambda}\mathcal{E}_{\mathbf{k}}^2$ . Why: the same surviving commutator as the previous card, now for every mode of free space at once.

Step 3. Convert to an integral and read off the divergence: with $\sum_{\mathbf{k}}\to V(2\pi)^{-3}\int d^3k$ , isotropy, $k=\omega/c$ , and two polarizations, $\langle 0|\hat{\mathbf{E}}^2|0\rangle=\int_0^\infty(\hbar\omega^3/2\pi^2\varepsilon_0 c^3)\,d\omega$ , and since electric and magnetic parts contribute equally the energy density is $u_0=\int_0^\infty\rho_0(\omega)\,d\omega$ with $\rho_0(\omega)=\hbar\omega^3/2\pi^2c^3$ ; the $\omega^3$ integrand has no cutoff, so $u_0\to\infty$ . Why: the $\omega^2$ density of states times $\hbar\omega$ per mode makes the integrand climb as $\omega^3$ — the vacuum's ultraviolet catastrophe, a real infinity signalling that only differences of $u_0$ reach observables. *Verified: $\rho_0(\omega)=\hbar\omega^3/2\pi^2c^3$ is the Planck spectral density with $\bar n\to\tfrac12$ ; consistent with Milonni.*

Go deeper · Derivation C · 1 minThe Casimir force from a difference of divergent sums

Derivation CThe Casimir force from a difference of divergent sums

The goal is to show how the same divergent zero-point sum, evaluated with and without conducting plates, yields a finite, measurable difference — the Casimir energy $E/A=-\pi^2\hbar c/720\,d^3$ and hence the attractive force. We count the modes a gap of width $d$ allows, subtract the free-space count under a regulator, and differentiate the energy to the pressure. The full result needs zeta-function regularization, but the logic — subtract two infinities to get an observable — is the whole physical idea.

Step 1. Count the allowed modes: between perfect conductors a distance $d$ apart the field along the gap is quantized, $k_z=n\pi/d$ for integer $n\ge0$ , while $\mathbf{k}_\perp$ stays continuous, so the enclosed energy per area sums $\tfrac{\hbar c}{2}\sqrt{k_\perp^2+(n\pi/d)^2}$ over $n$ . Why: shrinking $d$ pushes the rungs $n\pi/d$ apart, expelling low-frequency modes — the plates thin out the vacuum between them.

Step 2. Subtract free space under a regulator: the discrete sum minus its free-space integral, $\sum_n f(n)-\int_0^\infty f(n)\,dn$ , is finite by Euler–Maclaurin though each piece diverges, and leaves $E(d)/A=-\pi^2\hbar c/720\,d^3$ — negative because the gap holds less zero-point energy than free space. Why: real mirrors turn transparent at high frequency and cannot exclude arbitrarily short wavelengths, so the cutoff dependence cancels in the difference, leaving a result built only from $\hbar$ , $c$ , $d$ .

Step 3. Differentiate to the force: $P=-\tfrac{d}{dd}(E/A)=-\pi^2\hbar c/240\,d^4$ , so $F_{\text{Cas}}=PA=-\pi^2\hbar c\,A/240\,d^4$ . Why: force is the gradient of energy, and because the energy falls as $1/d^3$ the attraction steepens as $1/d^4$ — negligible at microns, enormous at nanometres. *Verified: $E/A=-\pi^2\hbar c/720\,d^3$ and $P=\pi^2\hbar c/240\,d^4$ ; the coefficient $\pi^2/240$ matches Casimir (1948) and Milonni.*

Limiting cases

Case	Result	Meaning
Single mode	$E_0=\tfrac12\hbar\omega$	finite, well-defined zero-point energy — the oscillator ground state
All free-space modes	$E_0=\sum_{\mathbf{k},\lambda}\tfrac12\hbar\omega_{\mathbf{k}}\to\infty$	the sum diverges; only regularized differences (Casimir) are physical
Spectral energy density	$\rho_0(\omega)=\dfrac{\hbar\omega^3}{2\pi^2c^3}$	zero-point density per frequency — Planck with $\bar n\to\tfrac12$ ; its $\omega^3$ integral is the UV catastrophe
High occupation	$\langle\hat{n}\rangle\gg1$	the $\tfrac12$ is negligible beside $\langle\hat{n}\rangle$ ; the field is effectively classical
Plate limit $d\to\infty$	$F_{\text{Cas}}\to0$	far apart, the gap holds the free-space spectrum; the force vanishes as $1/d^4$
Squeezed vacuum	$\langle\Delta\hat{X}_1^2\rangle<\tfrac14$	one quadrature dips below the vacuum floor, the other rises, product stays $\tfrac14$

Common trapThe zero-point energy of empty space is a definite, finite number you can quote

It is not. The absolute zero-point energy of free space is formally divergent, and no finite total energy density for the vacuum is an observable — never quote one as if measured. What physics measures is always a difference: how the vacuum energy changes when boundaries change (the Casimir force), or how a mode continuum shifts an atom's levels (the Lamb shift). Two traps ride along. First, $\langle 0|\hat{\mathbf{E}}|0\rangle=0$ does not mean the field is classically zero — the mean is zero but the variance $\langle 0|\hat{\mathbf{E}}^2|0\rangle$ is finite and does real work. Second, the Casimir force is not evidence for a definite vacuum energy density, only that changes in it are real — a weaker, correct statement.

Worked examples

1 · The Casimir pressure between two mirrors (the concrete one)

Two perfectly conducting plates face each other across a vacuum gap $d=1\ \mu\mathrm{m}$ . Find the Casimir pressure, the force on a $1\ \mathrm{cm}^2$ plate, and the pressure if the gap closes to $d=10\ \mathrm{nm}$ .

Step 1. Use $P=\pi^2\hbar c/240\,d^4$ with $\hbar=1.055\times10^{-34}\ \mathrm{J\,s}$ , $c=2.998\times10^8\ \mathrm{m/s}$ , $\pi^2=9.870$ . The numerator $\pi^2\hbar c=3.12\times10^{-25}\ \mathrm{J\,m}$ and $d^4=10^{-24}\ \mathrm{m}^4$ , so $P=3.12\times10^{-25}/(240\times10^{-24})=\mathbf{1.3\times10^{-3}\ Pa}$ — about 1.3 mPa, roughly ten billion times smaller than atmospheric.

Step 2. The force on a $1\ \mathrm{cm}^2=10^{-4}\ \mathrm{m}^2$ plate is $F=PA=1.3\times10^{-3}\times10^{-4}=\mathbf{1.3\times10^{-7}\ N}$ , about $0.13\ \mu\mathrm{N}$ — the scale torsion-balance experiments must resolve to see the effect.

Step 3. Scale to $d=10\ \mathrm{nm}$ : since $P\propto d^{-4}$ and the gap shrank by $100$ , the pressure grows by $100^4=10^8$ , giving $P=\mathbf{1.3\times10^{5}\ Pa}$ , about 1.3 atmospheres. The same vacuum imperceptible at a micron dominates mechanically at ten nanometres — the reason Casimir stiction plagues nanoelectromechanical devices. P = 1.3 mPa at 1 μm; F = 0.13 μN on 1 cm²; P ≈ 1.3 atm at 10 nm.

2 · The vacuum field amplitude of a cavity mode

A single optical mode at $\lambda=633\ \mathrm{nm}$ lives in a cavity of mode volume $V=1\ \mathrm{mm}^3$ . Estimate the rms vacuum field $\mathcal{E}_{\mathrm{vac}}=\sqrt{\hbar\omega/2\varepsilon_0 V}$ of that empty mode, then say what happens as the mode shrinks to a wavelength-sized volume.

Step 1. Angular frequency: $\omega=2\pi c/\lambda=2.98\times10^{15}\ \mathrm{rad/s}$ , so $\hbar\omega=3.14\times10^{-19}\ \mathrm{J}=1.96\ \mathrm{eV}$ , the photon energy of red light.

Step 2. Denominator $2\varepsilon_0 V=2\times8.854\times10^{-12}\times10^{-9}=1.77\times10^{-20}$ , so $\hbar\omega/2\varepsilon_0 V=17.7\ \mathrm{V^2/m^2}$ and $\mathcal{E}_{\mathrm{vac}}=\mathbf{4.2\ V/m}$ . Empty space in that mode is not electrically dead — it carries a jittering field of a few V/m.

Step 3. Shrink the mode: since $\mathcal{E}_{\mathrm{vac}}\propto V^{-1/2}$ , collapsing $V$ to a diffraction-limited $\lambda^3\approx2.5\times10^{-19}\ \mathrm{m}^3$ boosts the field by $\approx2\times10^{5}$ to $\mathcal{E}_{\mathrm{vac}}\approx\mathbf{2.6\times10^{5}\ V/m}$ . Tight confinement concentrates the vacuum fluctuations — the physics behind cavity QED and Purcell-enhanced emission. The corresponding zero-point energy $\tfrac12\hbar\omega=0.98\ \mathrm{eV}$ is unextractable, because $\hat{a}|0\rangle=0$ leaves no state below the vacuum to fall into. ≈ 4 V/m in 1 mm³, rising to ≈ 3×10⁵ V/m in a wavelength-sized mode; ½ħω = 0.98 eV.

Physics of the portrait

The vacuum's energy is a rare abstraction made to push on a real apparatus, and the person who saw how turned quantum electrodynamics into measurable mechanics.

Origins · 1 minHendrik Casimir and how this was discovered

Hendrik Casimir

1909–2000 · Dutch physicist who turned the vacuum's zero-point energy into a measurable force

Hendrik Brugt Gerhard Casimir studied under Paul Ehrenfest in Leiden and worked with Niels Bohr and Wolfgang Pauli before joining Philips, where he rose to head the research laboratories. In 1948, puzzling over why colloidal suspensions were more stable than van der Waals theory predicted, he and Dirk Polder reworked the attraction between neutral bodies to include the retardation of the electromagnetic interaction at large separations. Casimir then took the decisive leap: the same physics could be recast entirely in terms of the zero-point energy of the field, whose modes are reshaped by conducting boundaries. Two perfect mirrors, he showed, exclude the long-wavelength vacuum modes between them, lowering the enclosed energy relative to free space, so the energy gradient is an attractive pressure $P=\pi^2\hbar c/240\,d^4$ — built from nothing but $\hbar$ , $c$ , and the gap. Empty space, bounded by mirrors, pushes them together with a force made of fundamental constants alone. Early attempts by Marcus Sparnaay in the 1950s were consistent but crude; only in 1997 did Steve Lamoreaux, and soon after Umar Mohideen, confirm it to a few percent. It remains the cleanest demonstration that the quantum vacuum, though its absolute energy is infinite and unmeasurable, has differences that are utterly real.

History

1900

— Planck derives the blackbody radiation law, introducing energy quanta

1911

— Planck introduces a residual zero-point energy in his second radiation theory

1927

— Dirac quantizes the electromagnetic field, giving the photon and its vacuum

1947

— Lamb and Retherford measure the shift of hydrogen's 2S level against 2P

1948

— Casimir predicts an attractive force between two conducting plates

1997

— Lamoreaux measures the Casimir force, confirming it to a few percent

Going further

Texts. Loudon, The Quantum Theory of Light, 3rd ed., ch. 4 · Milonni, The Quantum Vacuum (1994), the definitive monograph · Scully and Zubairy, Quantum Optics, ch. 1–2 · Foot, Atomic Physics, ch. 7 · Saleh and Teich, Fundamentals of Photonics, 3rd ed., ch. 12.

Provenance. The single-mode results are standard quantum optics (Loudon ch. 4; Scully and Zubairy ch. 1). A verifier should recompute: (1) $\hat{H}=\hbar\omega(\hat{n}+\tfrac12)$ from $[\hat{a},\hat{a}^\dagger]=1$ ; (2) the vacuum variances $\langle\hat{X}_1^2\rangle=\langle\hat{X}_2^2\rangle=\tfrac14$ using $\hat{a}|0\rangle=0$ ; (3) the mode-sum-to-integral conversion giving $\rho_0(\omega)=\hbar\omega^3/2\pi^2c^3$ ; (4) the Casimir coefficients $\pi^2/240$ (pressure) and $\pi^2/720$ (energy) from Casimir 1948; (5) $P=1.3\ \mathrm{mPa}$ at $1\ \mu\mathrm{m}$ and $\mathcal{E}_{\mathrm{vac}}\approx4.2\ \mathrm{V/m}$ for $V=1\ \mathrm{mm}^3$ at $633\ \mathrm{nm}$ . The Lamb shift ( $\approx1.06\ \mathrm{GHz}$ for hydrogen $2S_{1/2}$ – $2P_{1/2}$ ) is the standard measured value (Foot ch. 7); the Casimir force was confirmed to a few percent by Lamoreaux (1997) and Mohideen (1998).

Prerequisites. Quantization of the Electromagnetic Field · Fock States and the Photon Number Basis · Quadrature Operators and the Uncertainty Relation. Leads to. Coherent States · Squeezed States of Light · Spontaneous Parametric Down-Conversion.

Ask the tutor: Ask me to show how the same $\tfrac12\hbar\omega$ that looks like a harmless constant becomes the Casimir force, the Lamb shift, and the shot-noise floor of LIGO.

Is the zero-point energy real, or just a bookkeeping constant we could subtract away?

It is real in the only sense physics allows: its differences do measurable work. You may redefine the energy zero and subtract the constant $\sum\tfrac12\hbar\omega$ in flat, unbounded space — nothing observable changes, since a uniform offset is invisible in non-gravitational physics. What you cannot subtract is the change when you insert mirrors, atoms, or a dielectric: the Casimir force and the Lamb shift are exactly such changes, and they are measured. The absolute value is convention; the vacuum's response to boundaries is a genuine feature of nature.

If the vacuum has energy everywhere, why doesn't it gravitate and wreck cosmology?

This is the crux of the cosmological-constant problem, and it is unsolved. Non-gravitational physics cares only about differences, so the divergent $\sum\tfrac12\hbar\omega$ is harmless there. But gravity couples to absolute energy density, so a naive vacuum energy should curve spacetime enormously: cutting the mode sum at the Planck frequency overshoots the observed dark-energy density by some 120 orders of magnitude, the largest known theory–observation mismatch. The resolution is open; quantum optics sidesteps it entirely by only ever using differences.

How is the vacuum field responsible for spontaneous emission?

An isolated excited atom in true vacuum has, classically, no field to stimulate it, so a semiclassical picture predicts it should sit excited forever. It does not — it decays at the Einstein $A$ -rate. The quantized field supplies the missing agent: vacuum fluctuations of the continuum of empty modes couple to the dipole and induce the transition. The rate follows from the density of vacuum modes the atom can emit into, which is why a cavity — reshaping that density — speeds up or suppresses emission. Spontaneous emission is, at bottom, emission stimulated by the vacuum.

The vacuum has ⟨E⟩ = 0 but ⟨E²⟩ ≠ 0. What does that actually mean physically?

It means the field averages to zero over time but is never actually zero at any instant — it fluctuates symmetrically about zero, like a fair coin (mean zero) that still always lands heads or tails (variance one). An ideal field of exactly zero would carry no energy; the finite second moment $\langle 0|\hat{\mathbf{E}}^2|0\rangle$ stores the $\tfrac12\hbar\omega$ per mode and jostles atoms. Whenever the vacuum does something measurable — driving emission, shifting a level, adding shot noise — it is the variance at work, never the mean. First moments describe classical fields; second moments describe quantum fluctuations.

Why does the Casimir force need a subtraction of two infinities to come out finite?

Because both the zero-point energy inside the gap and the energy that region would hold in free space are individually divergent. The meaningful quantity is their difference, the cost of inserting the plates, and that is finite. The regularization that makes it finite (a smooth cutoff, or zeta functions) is no trick: real mirrors turn transparent to X-rays, so they genuinely fail to exclude very short wavelengths, and the cutoff encodes that. When the regulator is removed its details cancel between the two sums, leaving a result built only from $\hbar$ , $c$ , and $d$ .

How does squeezing beat the vacuum noise floor without violating uncertainty?

The uncertainty relation constrains only the product $\Delta\hat{X}_1\,\Delta\hat{X}_2\ge\tfrac14$ , not each factor separately. The vacuum spends its allowance symmetrically, $\Delta\hat{X}_1=\Delta\hat{X}_2=\tfrac12$ ; a squeezed vacuum keeps the product at $\tfrac14$ but redistributes it, one quadrature dropping below $\tfrac14$ while the conjugate rises to compensate. If your measurement reads out only the squeezed quadrature — as a gravitational-wave interferometer can be arranged to do — you see noise below the vacuum floor, at the cost of extra noise in the quadrature you ignore. Uncertainty is never violated; the fluctuations are reshaped, not removed.